Three specialist agents judge flood conditions on a specified English river, each from its own evidence: forecasts, river gauges, flood records. Where they disagree, an orchestrator weighs them and explains which view should count for more, and why.
MeteorologyFour weather models and an ensemble (Open-Meteo)
HydrologyRiver levels, rain gauges, past events (Environment Agency). No forecast.
→
RiskTests Hydrology against gauge records, flood areas, history
→
OrchestratorSees only the three verdicts, and reconciles them
→
You seeEach verdict, then where they agree, conflict, and why
Replay a real event
the agents see only what was known at the time
Assess now
Live data from today. Each run makes around 15 model calls and takes about two minutes, so live runs are for invited guests. Have an invite link? Open it, or sign in at paulmartin.ai.
Or enter a postcode in England. It is matched to its nearest river gauge and rain gauges; only river flooding is assessed. It is used for this run only and not stored.
The agents never saw this. Reveal it to draw what really happened onto the chart below.
At a glance
what each agent saw and concluded
Daily flood risk is the orchestrator's assessment, on the Flood Forecasting Centre's scale:
✓ very low! low!! medium!!! high; – not rated where the agents gave no basis.
It is not an official forecast. Hover over any bar, line or day for details.